Srudies to develop technique and system which allow computers to accept speech inputs have been actively studied since the fifties. The natural question to ask is why study speech recognition. For practical reason speech recognition will solve problems, improve productivity and most important of all it will change the way we live today. As we improve algorithms and have faster machine, it appears that man-machine interface by voice will be a reality within our lifetime. In short term applciation spepech could be used to aid the handicapped (wheelchairs, robotic aid, control system, etc). A comparative study was made using different algorithms to cahiece the short term goal. the three models to be dexcribed are the LPC/DTW, LPC/DTW?VQ and the Neural Network. The fist two model used the template based approach. Distance measures are used to compare templates to find the best match. Dynamic programming is used to solve temporal difference. The technique of data compression is applied to one of these models. The other approach to speech recognition is the connectionist method. This is the most recent development in speech recognition. Connectionist apparocah consistes of many simple computing elements. Connection between these elements are of varying strength. The connection are trained to recognize speech. Statistical evaluation on a prototype system utilizing the recognition methjods mentioned above is as follows; The first model performs 95% recognition accuracy, the second model 92% accuracy and the connectionist model has 59% accuracy in normal quiet room.